1

On Reset of the Wind-Forced Decadal Kuroshio Extension2

Variability in Late 20173

Bo Qiu1, Shuiming Chen, Niklas Schneider4

Department of Oceanography, University of Hawaii at Manoa, Honolulu, Hawaii5

Eitarou Oka6

Atmosphere and Ocean Research Institute, The University of Tokyo, Kashiwa, Japan7

Shusaku Sugimoto8

Department of Geophysics, Graduate School of Science, Tohoku University, Sendai, Japan9

Journal of Climate10

Accepted: 30 September 202011

1Corresponding author address: Dr. Bo Qiu, Department of Oceanography, University of Hawaii at

Manoa, 1000 Pope Road, Honolulu, HI 96822. E-mail: bo@soest.hawaii.edu

2

Abstract12

Decadal modulations of the Kuroshio Extension (KE) system between a stable and13

an unstable dynamic state in the western North Pacific have prevailed in the past three14

decades. This dominance of decadal variations is controlled by the negative feedback loop15

involving the wind-forced KE variability and its feedback onto the overlying extratropical16

stormtracks and the basin-scale surface wind field. The wind-forced decadal KE modulations17

were disrupted in August 2017 due to the development of the Kuroshio large meander south18

of Japan. By forcing the inflow KE paths northward and by avoiding override the shallow19

Izu Ridge, the Kuroshio large meander was able to compel rapidly the KE from the wind-20

forced, pre-existing, unstable state to a stable state. Following the large meander occurrence21

in late 2017, the stabilized KE change is found to affect the overlying stormtracks and the22

basin-scale wind field the same way as those generated by the wind-forced KE change prior23

to 2017. Given the consistent atmospheric response to both the large-meander-induced and24

wind-forced KE variability, we expect that the KE dynamic state will resume its decadal25

modulation after the phase reset relating to the 2017 large meander event.26

3

1. Introduction27

The Kuroshio has its origin in the northern branch of the North Equatorial Current28

(NEC) that flows westward and bifurcates off the Philippine coast near 12◦N (Fig. 1). As it29

flows poleward, the Kuroshio gains progressively its volume transport and becomes a well-30

defined western boundary current (WBC) with a transport of ∼ 15 Sv when reaching the31

Luzon Strait (1 Sv = 106 ms−3; Lien et al. 2014). After passing by Taiwan, the Kuroshio32

enters the semi-enclosed East China Sea wherein its path is dynamically constrained by the33

steep continental slope. The Kuroshio enters the deep Shikoku basin through the Tokara34

Strait near 130◦E and, due to the development of an offshore recirculation gyre, its eastward35

volume transport can reach up to 60 Sv (e.g., Imawaki et al. 2001). The Kuroshio exits36

the Shikoku basin by overriding the shallow Izu Ridge along 140◦E and it is renamed the37

Kuroshio Extension (KE) after it enters the open North Pacific. With the enhancement from38

both the southern and northern recirculation gyres, the volume transport of the eastward-39

flowing KE has been observed to exceed 100 Sv (Wijffels et al. 1998; Qiu et al. 2008; Jayne40

et al. 2009).41

Large-amplitude variability in the Kuroshio/KE system can be separated into three42

segments, each with distinct regional characteristics. The first segment lies in the latitudinal43

band east of Luzon and Taiwan between 18◦–25◦N. As indicated in Fig. 1, this band has a44

high mesoscale eddy variability that extends eastward to the Hawaiian Islands across the45

western North Pacific Ocean. Mesoscale eddies along the 18◦–25◦N band have been observed46

to propagate westward with a typical period of O(100 days) and a phase speed of about47

0.1 m s−1, and they are generated by baroclinic instability induced by the zonal velocity48

shear between the eastward-flowing Subtropical Countercurrent (STCC) in the surface layer49

and the westward-flowing NEC in the subsurface layer in the western North Pacific (Qiu50

4

1999; Kobashi and Kawamura 2002; Cheng et al. 2017). After impinging upon the western51

boundary, the STCC mesoscale eddies have been detected to exert significant changes in52

the Kuroshio path and transport inside the East China Sea, as well as the Ryukyu Current53

that flows northward to the east of the Ryukyu Islands (e.g., Zhang et al. 2001; Gilson and54

Roemmich 2002; Ichikawa et al. 2004; Andres et al. 2008).55

The second segment of enhanced Kuroshio variability appears in the deep Shikoku basin56

south of Japan. Constrained by the inflow via the narrow Tokara Strait and the need to57

override the meridionally-oriented Izu Ridge along 140◦E, the Kuroshio in the Shikoku basin58

is known for its bimodal path fluctuations between a near-shore straight path and an offshore59

meandering path (Kawabe 1995). During both of these two paths (commonly known as the60

“straight path” and “large meander path”), the Kuroshio exits the Shikoku basin via a61

relatively deep channel centered around 34◦N along the shoaling Izu Ridge. In addition62

to these two preferred paths, the Kuroshio south of Japan can also exhibit a third path63

that loops southward over the shoaling Izu Ridge. The Kuroshio path fluctuations south of64

Japan are highly irregular and chaotic; the occurrence of the Kuroshio large meander (LM65

hereafter) shows no dominant periodicities, nor a preferred duration (for recent reviews, see66

Qiu and Miao 2000; Usui et al. 2013). As will be explored in section 3 of this study, the67

dynamical causes underlying the bimodal Kuroshio path fluctuations in the Shikoku basin68

remains to be a topic of active research.69

Free from the constraint of coastal boundaries, the KE east of Japan constitutes the70

third high eddy variability segment. With the enhancement of its volume transport and71

horizontal shear, the KE has long been observed to be an eastward-flowing inertial jet72

accompanied by large-amplitude meanders and energetic pinched-off eddies (e.g., Mizuno73

and White 1983; Yasuda et al. 1992). One salient feature emerging from recent satellite74

5

and eddy-resolving ocean general circulation model simulations is that the KE exhibits75

well-defined decadal modulations between a stable and an unstable dynamic state (Qiu and76

Chen 2005, 2010; Taguchi et al. 2007; Cebollas et al. 2009; Sugimoto and Hanawa 2009;77

Sasaki et al. 2013; Pierini 2014; Bishop et al. 2015). For example, Fig. 2 shows the KE78

paths were relatively stable in 1993–1994, 2002–2005, 2010–2015 and 2018–2019, based on79

merged satellite altimeter observations. In contrast, spatially convoluted paths prevailed80

during 1995–2001, 2006–2009, and 2016–2017. This decadal modulation of the KE system81

has persisted after the 1976–77 “climate shift” in the North Pacific (Qiu et al. 2014) and82

the basin-wide wind stress curl variability associated with the Pacific Decadal Oscillations83

(PDOs; Mantua et al. 1997), or the North Pacific Gyre Oscillations (NPGOs; Di Lorenzo et84

al. 2008), has been identified as the external forcing that controls the phase change between85

the stable/unstable dynamic states.86

Given their distinct characteristics, the oceanic circulation variability in these three87

segments of the North Pacific wind-driven WBC system has largely been explored sepa-88

rately in the past studies. The objective of our present study is to use the occurrence of89

the Kuroshio large meander in late 2017 to demonstrate that the variability in the afore-90

mentioned three segments are dynamically inter-connected. To better understand the roles91

of the KE variability upon the overlying atmosphere, we argue that it is, in fact, critical92

to clarify the interactions among the mesoscale eddies from the STCC band, the Kuroshio93

path fluctuations south of Japan, and the downstream KE variability in the open Pacific94

Ocean.95

This paper is organized as follows. After describing the observational datasets in section96

2, we examine in section 3 the KE dynamic state variability and its influence upon the97

overlying atmosphere. The negative feedback loop that enhances the decadal fluctuations98

6

in the extratropical ocean-atmosphere of the North Pacific is reviewed. In section 4, the99

bimodal Kuroshio path fluctuations in the Shikoku basin are examined with a focus on the100

on-going Kuroshio LM event that started in August 2017. In section 5, we address how101

the LM-induced KE dynamic state change affects the oceanic and overlying atmospheric102

circulation, and point to the reset of the negative feedback loop in the extratropical North103

Pacific ocean–atmosphere system by the 2017 LM occurrence. Discussion is provided in104

section 6 and section 7 summarizes the findings from the present study.105

2. Observational datasets106

To examine the surface oceanic circulation changes, we use the global sea surface height107

(SSH) dataset processed by Ssalto/Duacs and distributed by the Copernicus Marine and En-108

vironment Monitoring Service (CMEMS) (http://www.marine.copernicus.eu). This dataset109

merges along-track SSH measurements from all satellite altimeter missions after October110

1992 and has a 7-day temporal resolution and a 1/4◦ spatial resolution. The data period111

analyzed in this study extends from January 1993 to December 2019.112

To explore the subsurface circulation changes surrounding the KE jet, we utilize113

the conductivity-temperature-depth (CTD) surveys conducted by Japan Meteorological114

Agency (JMA) southeast of Japan on February 28–March 11, 2017 versus March 2–115

12, 2018. The CTD data has a vertical resolution of 1 db in the 2,000m upper ocean116

and can be accessed from https://www.data.jma.go.jp/gmd/kaiyou/db/vessel obs/data-117

report/html/ship/ship.php.118

To relate the oceanic variability to those at the air-sea interface and in the overlying119

atmosphere, we adopt the daily data from the European Centre for Medium–Range Weather120

7

Forecasts (ECMWF) ERA-5 reanalysis product (https://cds.climate.copernicus.eu/). The121

ECMWF ERA-5 data has a spatial resolution of 0.25◦ and is available from January 1979 to122

present. To infer the location/baroclinicity of extratropical storm-tracks, we follow Hoskins123

and Valdes (1990) and calculate the maximum Eady growth rate, σBI = 0.31fN−1∂U/∂z,124

in the lower troposphere. Specifically, the vertical zonal wind shear ∂U/∂z and the static125

stability parameter N are evaluated using the ERA-5 data between 700- and 850-hPa levels.126

To quantify the amplitude of extratropical storm-track activities, we follow Nakamura et al.127

(2002) and calculate the synoptic-scale (2–8 days) meridional transient eddy temperature128

fluxes < v′T ′ > from the hourly 850-hPa level ERA-5 data.129

3. KE dynamic state and its prediction/verification130

There are several metrics to quantify the decadal variability of the KE system depicted131

visually in Fig. 2. Figure 3a, for example, shows the KE path length integrated from 141◦E132

to 153◦E. A short (long) path length here corresponds to a stable (unstable) KE dynamic133

state. Similarly, Figs. 3b–3c show the surface eastward transport and latitudinal position of134

the KE averaged from 141◦E to 165◦E, and Fig. 3d shows the strength of the KE’s southern135

recirculation gyre based on the satellite altimeter SSH data of the past 27 years. In its stable136

dynamic state, the KE jet tends to have an intensified eastward transport, a northward137

latitudinal position, and an enhanced southern recirculation gyre. The reverse is true when138

the KE jet switches to an unstable dynamic state. Given the close connections among the139

dynamic properties shown in Fig. 3, Qiu et al. (2014) combined these four time series (by140

reversing the sign of KE path length and averaging the four time series normalized by their141

respective standard deviations) and introduced the “KE index” to concisely represent the142

8

low-frequency changes of the KE system. As shown in Fig. 4a, a positive (negative) KE index143

is defined to represent a stable (unstable) KE dynamic state. Statistically, the KE index is144

correlated favorably with the SSH anomaly signals averaged in the southern recirculation145

gyre region of 31◦–36◦N, 140◦–165◦E (Fig. 4b). This favorable correlation makes dynamical146

sense because a stable KE state is accompanied by an intensified southern recirculation147

gyre, hence a positive regional SSH anomaly.148

Like all other extratropical WBCs, the Kuroshio/KE transport a significant amount of149

warmer tropical water poleward, providing a source of heat and moisture for the midlatitude150

atmosphere (Kelly et al. 2010). This source of oceanic heat/moisture has been demonstrated151

by many recent studies to help maintain the lower tropospheric baroclinicity and anchor152

the extratropical stormtracks (Nakamura et al. 2004; Minobe et al. 2008; Kwon et al. 2010;153

Booth et al. 2012; Small et al. 2014; Masunaga et al. 2016; Ma et al. 2016; Bishop et al.154

2017; Parfitt and Seo 2018). Many data analysis studies in the past decades have examined155

the impact of the SST changes in the KE region on the atmospheric circulation across the156

midlatitude North Pacific basin (Frankignoul and Sennechael 2007; Qiu et al. 2007, 2017;157

Frankignoul et al. 2011; Taguchi et al. 2012; Smirnov et al. 2015; Ma et al. 2015; O’Reilly158

and Czaja 2014; Revelard et al. 2016). A consistent feature resulting from these analyses159

is that when the KE dynamic state is stable, increased surface turbulent (i.e., sensible +160

latent) heat fluxes tend to emit from ocean to atmosphere along the poleward-shifted KE161

path (Fig. 5a; defined positive from ocean to atmosphere). This enhanced turbulent heat162

flux forcing results in northward migration/amplification of the extratropical stormtracks163

as can be inferred from the lower tropospheric Eady growth rate and transient eddy <164

v′T ′ > distributions shown in Figs. 5b–5c. In connection with the meridional migration of165

stormtracks, the Ekman pumping velocity (wEk) field tends to similarly shift northward166

9

across the North Pacific basin (Fig. 5d).167

Notice that the time-mean stormtracks follow roughly the wEk = 0 line denoted by the168

thick black contour in Fig. 5. This line is tilted SW–NE across the North Pacific basin due to169

the combined orographic constraint and diabatic oceanic forcing (Wilson et al. 2009). When170

the stormtracks migrate northward during the KE’s stable state, a negative wEk anomaly171

band appears that straddles the time-mean wEk = 0 line (Fig. 5d). South of this band, the172

wEk anomalies turn positive because the northward migration of the wind system brings173

in stronger, subtropical-origin, negative wEk signals from the south. As a consequence, the174

anomalous wEk in the 31◦–36◦N band of the eastern North Pacific becomes positive. For175

the lower atmosphere, this positive wEk anomaly brings about enhanced regional rainfall176

over the eastern North Pacific (Ma et al. 2015). For the upper ocean, this leads to regional177

Ekman flux divergence and negative SSH anomalies in the eastern North Pacific Ocean.178

When these wind-forced negative SSH signals propagate westward into the KE region with179

a delay of 3∼ 4 years, they work to weaken the southern recirculation gyre, shifting the180

KE jet southward to override the shallow Izu Ridge, and transforming the KE system to181

an unstable dynamic state (e.g., Qiu and Chen 2005; Taguchi et al. 2007). Once the KE182

system switches to its unstable state, the reverse of the atmospheric responses depicted in183

Fig. 5 takes place and a dynamic state transition to a stable state occurs following a delayed184

oceanic adjustment. Although the KE-induced basin-scale atmosphere response is weak in185

comparison with the internal atmospheric fluctuations (typically at 10∼ 15% level for the186

interannual and decadal signals), this ocean-to-atmosphere feedback loop provides a nega-187

tive feedback mechanism that was argued by many recent studies to explain the enhanced188

decadal variability observed in the extratropical North Pacific ocean and atmosphere (Qiu189

et al. 2014; Smirnov et al. 2015; Na et al. 2018).190

10

It is relevant to mention that many of the studies cited above have related the KE’s191

dynamic state variability to the basin-wide wind forcing associated with the PDOs. As192

shown in Fig. 4d, there is a good correspondence between the negative PDO index and the193

KE index, or the recirculation gyre SSH signals, with the former leading the latter by 3 years194

and a correlation coefficient r = 0.52. Similar good correspondence also exists between the195

KE index and the NPGO index put forth by Di Lorenzo et al. (2008). Physically, this is196

not surprising because the KE dynamic state change is caused by the low-frequency surface197

wind stress fluctuations in the North Pacific basin that are encapsulated by leading climate198

indices, such as the PDO and the NPGO.199

The afore-mentioned negative feedback loop not only provides a mechanism for explain-200

ing the enhanced decadal variability, but also a framework for predicting the KE dynamic201

state. Relying on the slow oceanic baroclinic adjustment that carries the wind-forced SSH202

signals into the KE region and the basin-scale stormtracks and Ekman pumping field re-203

sponse to the changing KE dynamic state, we have attempted to forecast the SSH signals in204

the KE’s recirculation region or, equivalently, the KE index for the 2013–2022 period based205

on the observed SSH field of 2012 (see green line in Fig. 4c; replotted from Fig. 8 in Qiu et206

al. 2014). Compared to the observed SSH time series shown in Fig. 4b, it is interesting to207

note that the prediction did a good job in forecasting the positive-to-negative SSH change208

of the KE dynamic state in early 2017. The prediction, however, failed to forecast the209

abrupt reversal of the SSH signals to positive, i.e., stabilizing of the KE dynamic state, in210

late 2017.211

4. The 2017 Kuroshio large meander event212

11

What caused the “failure” of the predicted KE dynamic state in late 2017 based213

on the negative feedback loop between the KE and atmosphere stormtrack interactions?214

The answer is likely the occurrence of the Kuroshio LM that started in August 2017 in215

the upstream Shikoku basin. Figure 6a shows the monthly time series of the southern-216

most latitude of the Kuroshio south of Japan compiled by Japan Meteorological Agency217

based on combined 200m water temperature measurements and satellite observations218

(https://www.data.jma.go.jp/gmd/kaiyou/data/shindan/b 2/kuroshio stream/kuroshio219

stream.html). Shaded periods with latitude < 31.8◦N denote the LM state of the Kuroshio220

designated by JMA. As we noted in Introduction, the LM occurrences and durations are221

highly irregular. During the satellite altimeter era after 1992, one LM event took place in222

mid 2004 and lasted for 14 months. Following this short-lived event and a dormant period223

of 12 years, an intense and more persistent LM event initiated in August 2017 (Qiu 2019;224

Sugimoto et al. 2020) and the event is still on-going at the writing of this article.225

Due to the lack of persistent LM occurrence in the past three decades, the KE dynamic226

state, as shown in Fig. 4, has been dominated by the decadal variability whose timescale is227

dictated by the negative feedback loop between the KE and overlying atmospheric storm-228

track interactions. Indeed, this wind-forced KE variability was operative in the beginning229

of 2017 when the KE switched from a stable dynamic state to a negative one following the230

PDO phase change in early 2014 (Fig. 4d). After its dynamic state entered the unstable231

phase, Fig. 6b reveals that the Kuroshio path along 140◦E shifted southward over the shoal-232

ing Izu Ridge, resulting in fluctuating KE paths southeast of Japan (Qiu and Chen 2005,233

their section 5). Subsequent to the LM occurrence in August 2017, however, the offshore234

detouring Kuroshio becomes confined to the west of the Izu Ridge and its outflow is re-235

stricted to enter the open Pacific through a narrow, but deep (depth > 1, 100m), channel236

12

between the mainland Japan and offshore Izu Ridge (see Fig. 6c; Kawabe 1985). Forced by237

these changes in the upstream Kuroshio, the KE jet in the 140◦–153◦E segment becomes238

stabilized, leading to the reversal of the KE index in 2018–19 as indicated in Figs. 4a–4b.239

Given the capability of the Kuroshio path south of Japan in reversing the KE dynamic240

state, a question arising naturally is what led to the occurrence of the 2017 LM event. To241

address this question, it is instructive to examine the evolution of the SSH anomaly field242

prior to the occurrence of the LM event. As depicted in Fig. 7, the LM is well developed by243

September 2017 and its presence corresponds geostrophically to the negative SSH anomaly244

(indicated by dashed line in the figure) located to the south of Japan. Tracing it backwards245

in time, it is discernible that this negative SSH anomaly initiated in the upstream Kuroshio246

southeast of Kyushu during April–May 2017 and it can be further traced back to around247

145◦–150◦E in the 28◦–30◦N band as a sequence of cyclonic mesoscale eddies in October248

2016. These cyclonic eddies propagated westward at a speed of 0.11 m s−1 during November249

2016 to March 2017 and consolidated into a single negative SSH anomaly off Kyushu in250

April 2017.251

The negative SSH anomaly appearing in the upstream Kuroshio southeast of Kyushu in252

April–May 2017 is known as the “trigger” meander in existing literature and its importance253

as a precursor to the LM occurrence has been recognized in many previous studies (e.g.,254

Solomon 1978; Mitsudera et al. 2001; Ebuchi and Hanawa 2003; Miyazawa et al. 2004; Usui255

et al. 2008). In Fig. 8a, we plot the time series of the negative SSH anomalies averaged256

southeast of Kyushu within a 4◦ width from 29◦N to 32◦N (the green box in last panel of257

Fig. 7) based on the altimeter SSH data. The negative SSH anomaly triggering the 2017258

LM event corresponds to that appearing in early 2017. From this time series negative259

SSH anomalies, or trigger meanders, have appeared quite frequently off Kyushu during the260

13

past 3 decades. While a necessary condition, the trigger meander is clearly not a sufficient261

condition that will lead to a LM occurrence.262

A careful inspection of the sequential SSH anomaly maps of Fig. 7 reveals another factor263

that has contributed to the 2017 LM event. Following the trigger meander in May 2017,264

Fig. 7 reveals that the area off Kyushu is replaced by an intense positive SSH anomaly in265

July and August. This positive SSH anomaly appears to have two sources: one originated266

from the STCC band of 24◦–27◦N in October 2016 and moved subsequently northward to267

the east of the Ryukyu Islands; the other started around 153◦E in the 27◦–30◦N band in268

October 2016 and moved westward to merge with the STCC-originated anomaly off Kyushu269

in August 2017 (see the solid lines in Fig. 7). An important role played by this intense270

positive SSH anomaly is that it is able to thrust the preceding negative SSH anomaly, or271

the trigger meander, eastward, leading it to develop into a LM south of Japan.272

To examine if such positive SSH anomalies have succeeded other trigger meanders that273

appeared off Kyushu in the past, we construct in Fig. 8b the time series of the SSH anomalies274

averaged in the 4◦-width band east of the Ryukyu Islands from 24◦N to 29◦N (the green275

box in first panel of Fig. 7). As shown in Fig. 1, this band has an elevated SSH variance that276

connects the eddy-rich STCC to the Kuroshio southeast of Kyushu. The eddy variability277

in this band has positive and negative SSH anomalies with a timescale of typical oceanic278

mesoscales. To identify the events where a trigger meander southeast of Kyushu is followed279

by a positive SSH anomaly from east of the Ryukyu Islands, we multiply the SSH anomalies280

shown Fig. 8a to those of Fig. 8b with a lag of 4 months. Thus constructed, a negative281

product indicates an event in which a pre-existing trigger meander off Kyushu is succeeded282

by a positive SSH anomaly, with the magnitude of the product indicating their combined283

intensities. Figure 8c shows the time series of the above-mentioned negative product. Large284

14

negative products are now infrequent and they appeared in late 1998, late 2002, late 2003–285

early 2004, late 2006, late 2007, and early 2017, respectively. When compared to the time286

series of Kuroshio’s southernmost latitude south of Japan (black curve in Fig. 8c), it is287

of interest to note that the large negative products of late 2002 and late 2003–early 2004288

preceded the 2004–2005 LM event, and the early 2017 large negative product preceded the289

2017-present LM event. Although not categorized by JMA as LM events, the late 1998290

large negative product preceded the offshore meandering of the Kuroshio in 1999–2000, and291

the large negative products of late 2006 and late 2007 preceded the offshore meandering of292

the Kuroshio in 2008–2009.293

We have from the above analyses identified the combined emergence of trigger meander294

and subsequent positive SSH anomalies southeast of Kyushu as a prerequisite for the LM295

development south of Japan. It is worth emphasizing that the importance of anticyclonic296

perturbations in facilitating trigger meanders off Kyushu to develop into LM has been297

emphasized by a number of previous studies (Akitomo and Kurogi 2001; Kobashi and298

Hanawa 2004; Miyazawa et al. 2008; Usui et al. 2013; Tsujino et al. 2013). The origins299

of the anticyclonic perturbations proposed by these studies are, however, different from300

those identified in our analyses. All these previous studies have considered the anticyclonic301

perturbations of positive SSH anomalies to originate along the STCC band and be carried302

northward by the Kuroshio along the steep continental slope in the East China Sea. In light303

of the SSH anomaly maps shown in Fig. 7, we contend that the anticyclonic/positive SSH304

anomalies from the STCC band have more a capacity and a direct route east of the Ryukyu305

Islands, than deform the slope-constrained Kuroshio in the East China Sea, to affect the306

trigger meander that develops off Kyushu. This contention of ours is also supported by the307

satellite altimeter data (Fig. 1) revealing that an elevated SSH anomaly band exists to the308

15

east of the Ryukyu Islands that connects the eddy-rich STCC to the time-varying Kuroshio309

south of Japan.310

5. Impact on oceanic and atmospheric conditions311

In the preceding section, we have quantified the impact of the 2017 LM event on the312

KE dynamic state based on altimeter-measured SSH information. To assess further the313

KE change below the sea surface, we examine the in-situ CTD data collected by JMA314

in the upstream KE region before and after the 2017 LM occurrence. Figure 9 compares315

the temperature, salinity and cross-track geostrophic velocity profiles from JMA’s KS1702316

cruise of March 2017 versus those from the RF1802 cruise of March 2018 (see Figs. 6b–6c for317

locations of the CTD stations). From the T/S profiles, it is discernible that the KE front318

with the sharp T/S gradients in the 1,000m upper ocean shifted from 33.5◦N in March319

2017 to 34.5◦N in March 2018. Accompanied by this northward shift is an increase in320

KE’s transport and intensity: the maximum eastward geostrophic velocity and transport of321

the KE jet referenced to 2,000 db were 0.70 m s−1 and 70.2 Sv, respectively, from the 2017322

cruise (Fig. 9c) and these values increased to 0.88 m s−1 and 76.3 Sv during the 2018 cruise323

(Fig. 9f).324

Another noticeable difference between the two cruises is that the condition surrounding325

the KE jet was more variable in 2017 than in 2018. For example, Fig. 9a indicates the326

presence of an intense cold-core eddy centered around 30.5◦N in 2017. This cold-core eddy327

has a geostrophic transport of 50 Sv, close to 70% of the KE’s eastward transport value. Next328

to the coast of Japan, there appears a shallow branch of warmer/saltier subtropical-origin329

water in Figs. 9a–9b. A look at the sequential SSH maps (not shown) indicates that it was a330

16

remnant of the KE jet before its main body shifted southward in early 2017. Accompanying331

this southward shift of the KE jet, a fresh, subarctic-origin water mass with salinity < 34.0332

psu is seen to intrude to the north of the KE jet near 34◦N (Fig. 9b). In contrast to the 2017333

result, oceanic condition surrounding the KE jet was devoid of mesoscale perturbations in334

2018. The above results from the subsurface CTD measurements regarding the LM-induced335

changes in the KE’s position, intensity, and surrounding conditions, are all consistent with336

those of the preceding sections based on the altimeter SSH measurements.337

To evaluate if the 2017 LM-induced KE change exerted a similar impact upon the338

overlying atmosphere as that of the wind-forced dynamic state change in the past three339

decades, we plot in Fig. 10a the surface turbulent heat flux anomaly field averaged in the two340

years of 09/2017–08/2019 after the 2017 LM occurrence (this two-year window is selected341

because it is sufficiently long and contains two full seasonal cycles after the 2017 LM that342

are currently available to us). In conjunction with the northward shift and intensification343

of the KE jet, the surface turbulent heat fluxes exhibit positive anomalies (i.e., more heat344

supply from ocean to atmosphere) in the broad region east of Japan. This surface turbulent345

heat flux anomaly pattern is similar to that in Fig. 5a, showing the surface turbulent heat346

flux anomalies regressed to the KE index in 01/1979–07/2017 before the 2017 LM. In347

Figs. 10b–10c, we plot the stormtrack anomalies inferred from the lower tropospheric Eady348

growth rate and 850-hPa synoptic-scale transient eddy < v′T ′ > anomalies, in the two years349

after the 2017 LM occurrence. Consistent with the turbulent heat flux anomaly pattern350

shown in Fig. 10a, there exist an enhancement and a northward migration of the stormtrack351

activities in the recent two years. A comparison with Figs. 5b–5c indicates that a very352

similar stormtrack response to the KE change was detected in the 1979–2017 period prior353

to the 2017 LM event. Given the good agreement between the surface turbulent heat flux354

17

and stormtrack activity anomalies, it is of no surprise to find that a favorable correspondence355

also exists between the KE-induced wEk anomalies before and after the 2017 LM occurrence356

(cf. Figs. 10d and 5d). In short, the above comparisons indicate that the LM-induced KE357

variability after 2017 has exerted an impact upon the overlying atmospheric circulation358

similar to that by the wind-forced KE variability of the past three decades.359

6. Discussion360

Although the focus of our present study is on the recent interruption of the wind-361

forced decadal KE variability by the 2017 LM event, it is helpful to relate the KE dynamic362

state, the Kuroshio LM, and the basin-scale atmospheric variability from a longer timescale363

perspectives.364

a. Oceanic reach of the decadal KE variability365

In section 3, we combined four properties of the KE and defined the KE index to concisely366

describe the variations of the KE dynamic state. Although all these four properties are367

based on characteristics of the observed KE jet/recirculation gyre west of 165◦E, it does368

not imply that the oceanic influence of the decadal KE variability is confined regionally369

to the west of 165◦E. Figure 11a shows the altimeter-measured SSH anomaly distribution370

regressed to the KE index during the 1993–2019 period. With the anomalous SSH signals371

being a good proxy for the upper heat content anomalies (Kelly etal. 2010), it is clear from372

Fig. 11a that the positive heat content anomalies during the stable state state of KE extend373

both eastward beyond 165◦E and northeastward into the Oyashio Extension and subarctic374

frontal regions. This is so because as the KE flows eastward, it sheds mesoscale eddies and375

broadens its entity via northward bifurcations (Kida et al. 2015). In fact, our recent study376

18

reveals that the decadal Oyashio Extension variability along ∼ 40◦N and 153◦–173◦E is by377

and large dictated by the KE variability through northward shedded eddies and bifurcated378

branches (Qiu et al. 2017; their Fig.3). It is of interest to note that the spatial pattern of the379

turbulent heat flux anomalies regressed to the KE index (Fig. 5a) bears a close resemblance380

to that of Fig. 11a, lending the credence that it is the KE-originated variability that is381

responsible for the diabatic heating anomalies presented in Fig. 5a.382

b. Kuroshio LM and the PDO forcing383

That the Kuroshio LM paths are able to steer the low-level cyclone tracks away from the384

southern coast of Japan has been previously observed by Nakamura et al. (2012). In order385

to clarify if the Kuroshio LMs affect the atmospheric circulation across the broader Pacific386

basin, we define the Kuroshio path index as anomalies of the Kuroshio southernmost latitude387

away from the coast of Japan (Fig. 12a). Thus defined, a positive anomaly indicates a LM388

state of the Kuroshio south of Japan. Figure 11b shows the turbulent heat flux anomalies389

regressed to the Kuroshio path index with a two-month lag during the 1979–2019 period.390

Consistent with the findings by Nakamura et al. (2012), the diabatic heating effect of the391

LM events are confined to the regions surrounding Japan and there appears no systematic392

influence on the extratropical stormtracks across the North Pacific basin (figures not shown).393

This result indicates that although a LM event can alter the extratropical stormtracks by394

modifying the KE dynamic state as we found in section 5, changes by the Kuroshio paths395

alone may not be sufficient to induce the stormtrack variability across the North Pacific396

basin.397

We have emphasized in section 4 the roles played by mesoscale eddies along the 18◦–398

25◦N STCC band for the LM development. Previous studies have shown that the eddy399

activity level in the STCC band is modulated by the PDO-related wind forcing (Yoshida et400

19

al. 2011; Qiu and Chen 2013). Specifically, when the PDO is in positive phase, enhanced401

westerlies and trade winds over the western North Pacific generate a greater Ekman tem-402

perature flux convergence along the STCC band, causing a stronger upper ocean meridional403

temperature gradient and strengthening the vertically-sheared STCC/NEC system. This404

enhanced current shear results in a stronger baroclinic instability and elevates the level of405

regional eddy activities. A comparison between the Kuroshio path index and the PDO time406

series in Figs. 12a and 12b reveals that with the exception of the 1975 LM event, all other407

Kuroshio LM events were preceded by positive PDO phases. A lagged correlation analysis408

reveals that the Kuroshio path anomalies have a maximum correlation, r = 0.27, with the409

PDO index at a 32-month lag. Although not a high coefficient, this positive correlation410

is consistent with the notion that a positively-phased PDO forcing is able to enhance the411

mesoscale eddy activities in the STCC band, inducive for the formation of the Kuroshio412

LMs. The 32-month lag likely reflects the time required for (1) the baroclinic instability413

along the STCC to fully grow, (2) the instability-generated eddies to propagate westward414

across the STCC and poleward along the Ryukyu Island, and (3) the development from a415

trigger meander off Kyushu to the LM south of Japan.416

c. Imprint on PDO versus PMM by the KE variability417

Throughout this study, we have emphasized the negative feedback mechanism that in-418

volves the KE dynamic state, the meridional migration of extratropical stormtracks/gyre-419

scale Ekman pumping field, and the excitation of upper ocean baroclinic Rossby waves420

whose westward propagation alters the pre-existing KE dynamic state. Instead of invok-421

ing the stormtrack migration, Joh and Di Lorenzo (2019) have found recently that the422

decadally-varying KE can induce an eastern Pacific wind stress curl response that projects423

on atmospheric forcing of the Pacific Meridional Mode (PMM; Chiang and Vimont 2004).424

20

Specifically, when it is in a stable state, the KE-induced positive wEk anomaly shown in425

Fig. 5d in the 25◦–35◦N band of the eastern North Pacific projects to the positive PMM426

forcing pattern (Fig. 13a). This PMM forcing can activate the central tropical Pacific El427

Nino Southern Oscillation (CP-ENSO) that, in turn, can strengthen a positive PDO (or428

negative NPGO) forcing, generating the negative SSH anomalies in the eastern North Pa-429

cific and altering the KE to an unstable state after these wind-forced SSH anomalies reach430

the KE region.431

It is important to note that both our study and that of Joh and Di Lorenzo (2019) advo-432

cate the negative feedback mechanism that favors the decadal climate variability involving433

the KE system. The difference lies in whether the KE-induced positive wEk anomalies in the434

eastern North Pacific are a response confined to the midlatitude atmosphere via migration435

of stormtracks, or a response involving the tropical PMM/CP-ENSO variability. Figure 13b436

shows the spatial pattern of the wEk anomaly regressed to the PDO index. Similar to the437

PMM-regressed pattern shown in Fig. 13a, the PDO-regressed pattern shows a positive wEk438

anomaly in the eastern North Pacific that coincides with the positive wEk anomaly identified439

in Fig. 5d. In other words, like their projection on PMM, the KE-induced wEk anomalies440

project similarly on the PDO wind forcing pattern. The reason behind this similarity is441

because the PDO and PMM indices, as shown in Figs. 12b–12c, share similar time-varying442

signals especially on the decadal timescale of our interest (see also Stuecker 2018). Given443

these results, we believe it will be challenging to differentiate the negative feedback mech-444

anism described in this study and that proposed by Joh and Di Lorenzo (2019) from the445

statistical analyses alone. It will be crucial for future studies to clarify the atmospheric re-446

sponses to the KE variability based on process-oriented, coupled ocean-atmosphere general447

circulation model simulations.448

21

7. Summary449

Low-frequency Kuroshio Extension variability is composed of concurrent changes in its450

latitudinal position, eastward transport, eddy kinetic energy level, and southern recircu-451

lation gyre strength. In its stable dynamic state, the KE has a northward position, an452

increased transport, a reduced eddy kinetic energy level, and an intensified recirculation453

gyre, and the opposite is true during its unstable dynamic state. In the past three decades,454

the KE has observed to oscillate between these two dynamic states with a well-defined 10-455

year period. The dominance of this decadal oscillation is caused by the delayed negative456

feedback mechanism, in which the KE variability is driven by the basin-scale wind forcing457

with its center over the eastern North Pacific and the forced KE, in turn, modifies the458

overlying stormtracks and the large-scale wind pattern through anomalous turbulent heat459

fluxes from the ocean to atmosphere.460

The long-term, wind-forced KE variability was disrupted in August 2017 due to the461

occurrence of the Kuroshio large meander south of Japan. Instead of remaining in the462

unstable dynamic state that started in early 2017 in response to the 2014 phase change463

of the PDO wind forcing, the KE transitioned abruptly to a stable state in late 2017.464

The occurrence of LM is found from satellite altimeter measurements and available in-situ465

hydrographic surveys to impact the KE in two important ways. First, it forced the Kuroshio466

to enter the open North Pacific through a northerly deep channel close to Japan, causing467

the downstream KE path to shift northward. Second, by entering through the northerly468

deep channel, the Kuroshio avoided overriding the shoaling Izu Ridge and minimized the469

KE path fluctuations in the east of the Izu Ridge. Both of these effects worked to stabilize470

the KE dynamic state following the 2017 LM occurrence.471

Over the past three decades when the wind-forced KE variability prevails, the Kuroshio472

22

LM occurrence has been rare. The LM event preceding 2017 took place from July 2004 to473

August 2005 and this LM event exerted little impact upon the wind-forced KE variability474

because it occurred when the KE was already in a stable dynamic state. In agreement475

with the previous studies, we found that the LM occurrences in 2004–2005 and 2017–476

present were preceded by trigger meanders (represented by negative SSH anomalies) in the477

upstream Kuroshio southeast of Kyushu. For trigger meanders off Kyushu to develop fully478

into a LM state, we found it is necessary that they be succeeded by intense anticyclonic, or479

positive, SSH anomalies that work to thrust the trigger meander eastward and stationary480

south of Japan. Most of these anticyclonic/positive SSH anomalies have their origins in481

the STCC band of 18◦–25◦N and, rather than being advected by the Kuroshio through the482

East China Sea, they are observed to propagate northward to the east along the Ryukyu483

Islands.484

The LM-induced KE dynamic state change after late 2017 is found to affect the overlying485

stormtracks and the large-scale wind pattern the same way as those generated by the wind-486

forced KE change prior to the 2017 LM occurrence. This consistent atmospheric response487

implies that although disrupted temporally, the negative feedback loop and the decadally-488

modulating KE variability will likely resume once the current LM event terminates. It489

will be interesting to find out how long the current LM event will last and how the KE490

dynamic state will respond in the coming years to the LM-induced positive Ekman pumping491

anomalies in the 32◦–36◦N band over the eastern North Pacific Ocean.492

23

Acknowledgments493

This study was conducted while EO and SS were on their sabbatical visit to University494

of Hawaii. Constructive comments made by four anonymous reviewers have significantly495

improved an early version of the manuscript. The ECMWF ERA-5 reanalysis data used496

in this study was accessed from https://cds.climate.copernicus.eu/, the JMA hydrographic497

cruise data and analysis information from http://www.jma.go.jp/jma/indexe.html, and the498

merged satellite altimeter data from Ssalto/Duacs and the Copernicus Marine and En-499

vironment Monitoring Service (http://www.marine.copernicus.eu). We acknowledge the500

support of NSF grant 2019312 and NASA grant NNX17AH33G to BQ and SC, MEXT501

grant 19H05700 and JSPS grant 17K05652 to EO, and MEXT grant 19H05704 and JSPS502

grant 18K03737 to SS.503

References504

Akitomo, K., and M. Kurogi, 2001: Path transition of the Kuroshio due to mesoscale505

eddies: A two-layer, wind-driven wind-driven experiment. J. Oceanogr., 57, 735–741.506

Andres, M., J. H. Park, M. Wimbush, X. H. Zhu, K. I. Chang, and H. Ichikawa, 2008:507

Study of the Kuroshio/Ryukyu Current System based on satellite-altimeter and in508

situ measurements. J. Oceanogr., 64, 937–950.509

Bishop, S.P., F.O. Bryan, and R.J. Small, 2015: Bjerknes-like compensation in the win-510

tertime north Pacific. J. Phys. Oceanogr., 45, 1339–1355.511

24

——, R.J. Small, F.O. Bryan, and R.A. Tomas, 2017: Scale dependence of mid-latitude512

air-sea interaction. J. Climate, 30, 8207–8221.513

Booth, J.F., L. Thompson, J. Patoux, and K.A. Kelly, 2012: Sensitivity of midlatitude514

storm intensification to perturbations in the sea surface temperature near the Gulf515

Stream. Mon. Wea. Rev., 140, 1241–1256.516

Ceballos, L., E. Di Lorenzo, C. D. Hoyos, N. Schneider, and B. Taguchi, 2009: North517

Pacific Gyre Oscillation synchronizes climate variability in the eastern and western518

boundary current systems. J. Climate, 22, 5163–5174.519

Cheng, Y.-H., C.-R. Ho, Q. Zheng, B. Qiu, J. Hu, and N.-J. Kuo, 2017: Statistical fea-520

tures of eddies approaching the Kuroshio east of Taiwan Island and Luzon Island. J.521

Oceanogr., 73, 427–438.522

Chiang, J.C.H., and D.J. Vimont, 2004: Analagous meridional modes of atmosphere-ocean523

variability in the tropical Pacific and tropical Atlantic. J. Climate, 17, 4143–4158.524

Di Lorenzo, E., N. Schneider, K.M. Cobb, P.J.S. Franks, K. Chhak, A.J. Miller, J.C.525

McWilliams, S.J. Bograd, H. Arango, E. Curchitser, T.M. Powell, and P. Riviere, 2008:526

North Pacific gyre oscillation links ocean climate and ecosystem change. Geophys.527

Res. Lett., 35, L08607. https://doi.org/10.1029/2007gl032838.528

Ebuchi, N., and K. Hanawa, 2003: Influence of mesoscale eddies on variations of the529

Kuroshio paths south of Japan. J. Oceanogr., 59, 25–36.530

Frankignoul, C., and N. Sennechael, 2007: Observed influence of North Pacific SST anoma-531

lies on the atmospheric circulation. J. Climate, 20, 592–606.532

25

——, ——, Y.-O. Kwon, and M.A. Alexander, 2011: Influence of the meridional shifts of533

the Kuroshio and the Oyashio Extensions on the atmospheric circulation. J. Climate,534

24, 762–777.535

Gilson, J., and D. Roemmich, 2002: Mean and temporal variability in Kuroshio geostrophic536

transport south of Taiwan (1993–2001). J. Oceanogr., 58, 183–195.537

Hoskins, B.J., and P.J. Valdes, 1990: On the existence of storm-tracks. J. Atmos. Sci.,538

47, 1854–1864.539

Ichikawa, H., H. Nakamura, A. Nishina, and M. Higashi, 2004: Variability of north-540

eastward current southeast of northern Ryukyu Islands. J. Oceanogr., 60, 351–363.541

Imawaki, S., A.S. Bower, L. Beal, and B. Qiu, 2013: Western Boundary Currents. In542

Ocean Circulation and Climate - A 21st Century Perspective, 2nd Edition, G. Siedler,543

S.M. Griffies, W.J. Gould and J. Church (eds.); Academic Press, pp. 305–338.544

Jayne, S., N. Hogg, S. Waterman, L. Rainville, K. Donahue, D. Watts, K. Tracey, J.545

McClean, M. Maltrud, B. Qiu, S. Chen, and P. Hacker, 2009: The Kuroshio Extension546

and its recirculation gyres. Deep-Sea Res., 56, 2088–2099.547

Joh, Y., and E. Di Lorenzo, 2019: Interactions between Kuroshio Extension and Central548

Tropical Pacific lead to preferred decadal timescale oscillations in Pacific climate. Sci.549

Report, 9, 13558.550

Kawabe, M., 1985: Sea level variations at the Izu Islands and typical stable paths of the551

Kuroshio. J. Oceanogr. Soc. Jpn., 41, 307–326.552

26

——, 1995: Variations of current path, velocity, and volume transport of the Kuroshio in553

relation with the large meander. J. Phys. Oceanogr., 25, 3103–3117.554

Kelly, K.A., R.J. Small, R.M. Samelson, B. Qiu, T.M. Joyce, Y.-O. Kwon and M.F. Cronin,555

2010: Western boundary currents and frontal air-sea interaction: Gulf Stream and556

Kuroshio Extension. J. Climate, 23, 5644–5667.557

Kida, S., and Coauthors, 2015: Oceanic fronts and jets around Japan: A review. J.558

Oceanogr., 71, 469–497.559

Kobashi, F., and H. Kawamura, 2002: Seasonal variation and instability nature of the560

North Pacific Subtropical Countercurrent and the Hawaiian Lee Countercurrent. J.561

Geophys. Res., 107, C03185, doi:10.1029/2001JC001225.562

——, and K. Hanawa, 2004: Hydrographic features off the southeast coast of Kyushu563

during the Kuroshio small meanders: a case study for small meanders that occurred564

in 1994 and 1995. J. Oceanogr., 60, 645–661.565

Kwon, Y.-O., M.A. Alenxader, N.A. Bond, C. Frankignoul, H. Nakamura, B. Qiu, and L.566

Thompson, 2010: Role of the Gulf Stream and Kuroshio-Oyashio systems in large-567

scale atmosphere-ocean interaction: A review. J. Climate, 23, 3249–3281.568

Lien, R.-C., B. Ma, Y.-H. Cheng, C.-R. Ho, B. Qiu, C.M. Lee, and M.-H. Chang, 2014:569

Modulation of Kuroshio transport by mesoscale eddies at the Luzon Strait entrance.570

J. Geophys. Res., 119, 2129-2142, doi:10.1002/2013JC009548.571

Ma, X., P. Chang, R. Saravanan, R. Montuoro, J.-S. Hsieh, D. Wu, X. Lin, L. Wu, Z.572

Jing, 2015: Distant influence of Kuroshio eddies on North Pacific weather patterns.573

Sci. Rep., 5, 17785.574

27

——, J. Zhao, P. Chang, X. Liu, R. Montuoro, R.J. Small, F.O. Bryan, R.J. Greatbatch,575

P. Brandt, D. Wu, X. Lin, and L. Wu, 2016: Western boundary currents regulated by576

interaction between ocean eddies and the atmosphere. Nature, 535, 533–537.577

Mantua, N.J., S.R. Hare, Y. Zhang, J.M. Wallace, and R.C. Francis, 1997: A Pacific578

interdecadal climate oscillation with impacts on salmon production. Bull. Amer.579

Meteor. Soc., 78, 1069–1079.580

Masunaga, R., H. Nakamura, T. Miyasaka, K. Nishii, and B. Qiu, 2016: Interannual581

modulations of oceanic imprints on the wintertime atmospheric boundary layer under582

the changing dynamical regimes of the Kuroshio Extension. J. Climate, 29, 3273–583

3296.584

Minobe, S., A. Kuwano-Yoshida, N. Komori, S.-P. Xie, and R. J. Small, 2008: Influence585

of the Gulf Stream on the troposphere. Nature, 452, 206–209.586

Mitsudera, H., T. Waseda, T. Yoshikawa, and B. Taguchi, 2001: Roles of mesoscale eddies587

in the Kuroshio paths. Geophys. Res. Lett., 28, 2025–2028.588

Miyazawa, Y., X. Guo, and T. Yamagata, 2004: Roles of mesoscale eddies in the Kuroshio589

paths. J. Phys. Oceanogr., 34, 2203–2222.590

——, T. Kagimoto, X. Guo, and H. Sakuma, 2008: The Kuroshio large meander formation591

in 2004 analyzed by an eddy-resolving ocean forecast system. J. Geophys. Res., 113,592

C10015, doi:10.1029/2007JC004226.593

Mizuno, K., and W.B. White, 1983: Annual and interannual variability in the Kuroshio594

Current system. J. Phys. Oceanogr., 13, 1847–1867.595

28

Na, H., K.-Y. Kim, S. Minobe, and Y.N. Sasaki, 2018: Interannual to decadal variability596

of the upper-ocean heat content in the western North Pacific and its relationship to597

oceanic and atmospheric variability. J. Climate, 31, 5107–5125.598

Nakamura, H., T. Sampe, Y. Tanimoto, and A. Shimpo, 2004: Observed associations599

among storm tracks, jet streams and midlatitude oceanic fronts. Earth’s Climate:600

The Ocean-Atmosphere Interaction. Geophys. Monogr., 147, Amer. Geophys. Union,601

329–346.602

——, T. Izumi, and T. Sampe, 2002: Interannual and decadal modulations recently ob-603

served in the Pacific storm track activity and East Asian winter monsoon. J. Climate,604

15, 1855–1874.605

Nakamura, H., A. Nishina, and S. Minobe, 2012: Response of storm tracks to bimodal606

Kuroshio path states south of Japan. J. Climate, 25, 7772–7779.607

O’Reilly, C.H., and A. Czaja, 2014: The response of the Pacific storm track and atmo-608

spheric circulation to Kuroshio Extension variability. Quart. J. Royal Meteoro. Soc.,609

doi:10.1002/qj.2334.610

Parfitt, R., and H. Seo, 2018: A new framework for near-surface wind convergence over the611

Kuroshio Extension and Gulf Stream in wintertime: The role of atmospheric fronts.612

Geophys. Res. Lett., 45, https://doi.org/10.1029/2018GL080135.613

Pierini, S., 2014: Kuroshio Extension bimodality and the North Pacific Oscillation: A case614

of intrinsic variability paced by external forcing. J. Climate, 27, 448–454.615

Qiu, B., 1999: Seasonal eddy field modulation of the North Pacific Subtropical Coun-616

tercurrent: TOPEX/POSEIDON observations and theory. J. Phys. Oceanogr., 29,617

29

2471–2486.618

——, 2019: Kuroshio and Oyashio Currents. Encyclopedia of Ocean Sciences (3rd Edn),619

edited by J.K. Cochran, H. Bokuniewicz, P. Yager. Academic Press, 384–394.620

——, and S. Chen, 2005: Variability of the Kuroshio Extension jet, recirculation gyre and621

mesoscale eddies on decadal timescales. J. Phys. Oceanogr. 35, 2090–2103.622

——, and ——, 2010: Eddy-mean flow interaction in the decadally-modulating Kuroshio623

Extension system. Deep-Sea Res. II, 57, 1098–1110.624

——, and ——, 2013: Concurrent decadal mesoscale eddy modulations in the western625

North Pacific subtropical gyre. J. Phys. Oceanogr., 43, 344–358.626

——, and W. Miao, 2000: Kuroshio path variations south of Japan: Bimodality as a627

self-sustained internal oscillation. J. Phys. Oceanogr., 30, 2124–2137.628

——, S. Chen, N. Schneider, and B. Taguchi, 2014: A coupled decadal prediction of the629

dynamic state of the Kuroshio Extension system. J. Climate, 27, 1751–1764.630

——, ——, P. Hacker, N. Hogg, S. Jayne, and H. Sasaki, 2008: The Kuroshio Extension631

northern recirculation gyre: Profiling float measurements and forcing mechanism. J.632

Phys. Oceanogr., 38, 1764–1779.633

——, ——, and N. Schneider, 2017: Dynamical links between the decadal variability of634

the Oyashio and Kuroshio Extensions. J. Climate, 30, 9591–9605.635

——, N. Schneider, and S. Chen, 2007: Coupled decadal variability in the North Pacific:636

An observationally-constrained idealized model. J. Climate, 20, 3602–3620.637

30

Revelard, A., C. Frankignoul, N. Sennechael, Y.-O. Kwon, and B. Qiu, 2016: Influence of638

the decadal variability of the Kuroshio Extension on the atmospheric circulation in639

the cold season. J. Climate, 29, 2123–2144.640

Rio, M.-H., S. Guinehut, and G. Larnico, 2011: New CNES-CLS09 global mean dynamic641

topography computed from the combination of GRACE data, altimetry, and in situ642

measurements. J. Geophys. Res., 116, C07018, doi:10.1029/2010/JC006505.643

Sasaki, Y. N., S. Minobe, and N. Schneider, 2013: Decadal response of the Kuroshio644

Extension jet to Rossby waves: Observation and thin-jet theory. J. Phys. Oceanogr.,645

43, 442–456.646

Small, R.J., R.A. Tomas, F.O. Bryan, 2014: Storm track response to ocean fronts in a647

global high-resolution climate model. Clim. Dyn., 43, 805–828.648

Smirnov, D., M. Newman, M.A. Alexander, Y.-O. Kwon, and C. Frankignoul, 2015: Inves-649

tigating the local atmospheric response to a realistic shift in the Oyashio sea surface650

temperature front. J. Climate, 28, 1126–1147.651

Solomon, H., 1978: Occurrence of small “trigger” meanders in the Kuroshio off southern652

Kyushu. J. Oneanogr. Soc. Japan, 34, 81–84.653

Stuecker, M.F., 2018: Revisiting the Pacific Meridional Mode. Sci. Report, 8, 3216.654

Sugimoto, S., and K. Hanawa, 2009: Decadal and interdecadal variations of the Aleutian655

Low cctivity and their relation to upper oceanic variations over the North Pacific. J.656

Meteorol. Soc. Japan, 87, 601–614.657

——, B. Qiu, and A. Kojima, 2020: Marked coastal warming off Tokai attributable658

31

to Kuroshio large meander. J. Oceanogr., 76, https://doi.org/10.1007/s10872-019-659

00531-8.660

Taguchi, B., S.-P. Xie, N. Schneider, M. Nonaka, H. Sasaki, Y. Sasai, 2007: Decadal661

variability of the Kuroshio Extension: Observations and an eddy-resolving model662

hindcast. J. Climate, 20, 2357–2377.663

——, H. Nakamura, M. Nonaka, N. Komori, A. Kuwano-Yoshida, K. Takaya, and A. Goto,664

2012: Seasonal evolutions of atmospheric response to decadal SST anomalies in the665

North Pacific subarctic frontal zone: Observations and a coupled model simulation.666

J. Climate, 25, 111–139.667

Tsujino, H., S. Nishikawa, K. Sakamoto, N. Usui, H. Nakano, and G. Yamanaka, 2013:668

Effects of large-scale wind on the Kuroshio path south of Japan in a 60-year historical669

OGCM simulation. Climate Dyn., 41, 2287–2318.670

Usui, N., H. Tsujino, H. Nakano, and Y. Fujii, 2008: Formation process of the Kuroshio671

large meander in 2004. J. Geophys. Res., 113, C08047. doi:10.1029/2007JC004675.672

——, ——, ——, and S. Matsumoto, 2013: Long-term variability of the Kuroshio path673

south of Japan. J. Oceanogr., 69, 647–670.674

Wijffels, S. E., M. M. Hall, T. Joyce, D. J. Torres, P. Hacker, and E. Firing, 1998: Multiple675

deep gyres of the western North Pacific: A WOCE section along 149◦E. J. Geophys.676

Res., 103, 12,985–13,009.677

Wilson, C., B. Sinha, and R.G. Williams, 2009: The effect of ocean dynamics and orogra-678

phy on atmospheric storm tracks. J. Climate, 22, 3689–3702.679

32

Yasuda, I., K. Okuda, and M. Hirai, 1992. Evolution of a Kuroshio warm-core ring –680

Variability of the hydrographic structure. Deep–Sea Res., 39, 131–161.681

Yoshida, S., B. Qiu, and P. Hacker, 2011: Low-frequency eddy modulations in the Hawaiian682

Lee Countercurrent: Observations and connection to the Pacific Decadal Oscillation.683

J. Geophys. Res., 116, C12009, doi:10.1029/2011JC007286.684

Zhang, D., Lee, T.N., Johns, W.E., Liu, C.-T., Zantopp, R., 2001: The Kuroshio east of685

Taiwan: modes of variability and relationship to interior ocean mesoscale eddies. J.686

Phys. Oceanogr., 31, 1054–1074.687

33

rms SSHa [cm]0 2 4 6 8 10 12

20

20

20

30

30

30

40

40

40

40

50

50

50

50

50

60

60

6060

60

70

70 70

70 70 70

80

8080 80

80

80

90

9090

90

90

90

90

90

90

100

100

100

100100

100

100

100100

100

100

100

110

110

110 110

110

110

110

110

110

110

110

120

120

120

120

120

120

120120

130

130

130

130

130

130

130

140

140

140

140

140

140

150160170

120°E 130°E 140°E 150°E 160°E 170°E 180° 170°W 160°W 150°W5°N

10°N

15°N

20°N

25°N

30°N

35°N

40°N

45°N

Figure 1: Root-mean-squared sea surface height variability in the western North Pacificbased on high-pass filtered satellite altimeter data from January 1992 to February 2020.The high-pass filter has a half-power at 180 days. Regions where the rms variability exceeds12 cm are indicated by black contours with an interval of 2 cm. White contours denote themean SSH field by Rio et al. (2011).

34

28°N

32°N

36°N

40°N

1993

28°N

32°N

36°N

40°N

1994

28°N

32°N

36°N

40°N

1995

28°N

32°N

36°N

40°N

1996

28°N

32°N

36°N

40°N

1997

28°N

32°N

36°N

40°N

1998

140°E 150°E 160°E28°N

32°N

36°N

40°N

1999

2000

2001

2002

2003

2004

2005

140°E 150°E 160°E

2006

2007

2008

2009

2010

2011

2012

140°E 150°E 160°E

2013

2014

2015

2016

2017

2018

140°E 150°E 160°E

2019

Figure 2: Yearly paths of the Kuroshio and KE plotted every 14 days since 1993 based onsatellite altimetry observations (for detailed path derivation, see Qiu and Chen 2005). Red(blue) labels denote years in which the KE is largely in a stable (unstable) dynamic state.

35

93 95 97 99 01 03 05 07 09 11 13 15 17 19 0

0.5

1

1.5

2

S(t

) [1

012 m

3]

(d) KE Recirculation Gyre Strength

93 95 97 99 01 03 05 07 09 11 13 15 17 19

33°N

34°N

35°N

36°N

37°N(c) Upstream KE Position (141

o−165

oE)

93 95 97 99 01 03 05 07 09 11 13 15 17 19 10

30

50

70

δ h [cm

]

(b) KE Strength (141o−165

oE)

93 95 97 99 01 03 05 07 09 11 13 15 17 19 0

2000

4000

Path

Length

[km

]

(a) Upstream KE Path Length (141o−153

oE)

Figure 3: Time series of (a) upstream KE path length integrated from 141◦E to 153◦E, (b)SSH difference across the KE jet averaged from 141◦E to 165◦E, (c) latitudinal positionof the KE averaged from 141◦E to 165◦E, and (d) intensity of the KE recirculation gyre.Thick black lines denote the periods when the KE is in an unstable dynamic state. Formore details about the time series, see Qiu and Chen (2005).

36

−3

−1

1

3

KE

Index

(a)

−20

−10 0

10

20

SS

Ha [cm

]

(b)

−20

−10 0

10

20

SS

Ha [cm

]

(c)

1990 1995 2000 2005 2010 2015 2020−3

−1

1

3

Negative P

DO

(d)

Figure 4: (a) Time series of the KE index synthesized from the KE dynamical propertiesshown in Figs. 3a–3d. (b) Time series of SSH anomalies averaged in the KE southern recircu-lation region of 31◦–36◦N, 140◦–165◦E. (c) As in (b), but for the SSH anomalies of 2013-2019(the green line) predicted based on the observed SSH data of 2012. See Qiu et al. (2014)for derivations of the SSH anomaly time series and its prediction. (d) Time series of thenegative PDO index (available from http://research.jisao.washington.edu/data sets/pdo).Anomalies in (b)–(d) are relative to their respective 1989–2019 means and thick black linesin (a), (b) and (d) denote the time series after a low-pass filter that has a half-power at 6months.

37

−100 −50 0 50 100

(a)

Q’ Regression [W/m2/m]

ECMWF ERA5: 01/1979−07/2017

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

−0.4 −0.2 0 0.2 0.4

(b)

σBI

Regression [1/day/m]

140°E 160°E 180° 160°W 140°W 120°W

−6 −4 −2 0 2 4 6

(c)

v’T’ Regression [oCm/s/m]

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

−4 −2 0 2 4

(d)

Ekman Pumping Regression [10−6

m/s/m]

140°E 160°E 180° 160°W 140°W 120°W

Figure 5: (a) Regression coefficient between the surface turbulent heat flux anomaly fieldand the KE index of 01/1979–07/2017 with the former lagging the latter by 2 months. (b)As in (a), but for the maximum Eady growth rate anomaly field at 700–850 hPa level. (c)As in (a), but for the 2∼ 8-day band-passed meridional transient eddy temperature fluxanomaly field at 850-hPa level. (d) As in (a), but for the wEk anomaly field. In all figures,the thick black contour denotes the time-mean wEk = 0 line and stippled areas indicatewhere the statistical significance exceeds the 90% confidence level based on Monte Carlosimulations (for detailed methodology, see Qiu et al. 2007). Dashed box in (d) indicatesthe 31◦–36◦N band where the Ekman pumping forcing impacts the KE dynamic state.

38

1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 202029°N

30°N

31°N

32°N

33°N

34°N(a) Southernmost Latitude of Kuroshio

(b) Kuroshio Paths in 01/2017−07/2017

130°E 135°E 140°E 145°E 150°E28°N

30°N

32°N

34°N

36°N

38°N

40°N

(c) LM Paths in 09/2017−03/2018

130°E 135°E 140°E 145°E 150°E 155°E

Figure 6: (a) Time series of the monthly southernmost latitude of the Kuroshio south ofJapan compiled by Japan Meteorological Agency. Solid thick line shows the low-pass filteredtime series after a 13-month running-mean average. Shaded periods with latitude < 31.8◦Ndenote the large meander state of the Kuroshio. (b) Weekly Kuroshio/KE paths off Japanduring 01/2017 to 07/2017. Black dots denote CTD stations by JMA cruise KS17027. (c)Same as (b), but during 09/2017 to 03/2018 and JMA cruise RF1802. Green (grey) shadedenotes areas shallower than 500 (1500) m.

39

10/2016−09/2017 Monthly SSHa

Oct/16

20°N

24°N

28°N

32°N

36°N Nov/16 Dec/16

Jan/17

20°N

24°N

28°N

32°N

Feb/17 Mar/17

Apr/17

20°N

24°N

28°N

32°N

May/17 Jun/17

Jul/17

120°E 130°E 140°E 150°E 20°N

24°N

28°N

32°N

Aug/17

120°E 130°E 140°E 150°E

−40 −30 −20 −10 0 10 20 30 40 SSHa [cm]

Sep/17

120°E 130°E 140°E 150°E

Figure 7: Monthly maps of SSH anomalies from October 2016 to September 2017 in thewestern North Pacific Ocean. To avoid large-scale steric height changes due to seasonalheating/cooling, the area mean SSH anomaly value of each month is removed. Dashedlines denote the evolution of the trigger meander that appeared southeast of Kyushu inApril 2017 and solid lines, the positive SSH anomalies that replaced the trigger meanderoff Kyushu in August 2017. The green box in the first (last) panel denotes the area usedto construct the SSH anomaly time series shown in Fig. 8b (8a).

40

1995 2000 2005 2010 2015 2020−40

−30

−20

−10

0

h’ k

[cm

](a) h’

k (< 0 only)

1995 2000 2005 2010 2015 2020−20

−10

0

10

20

h’ o

[cm

]

(b) h’o

1995 2000 2005 2010 2015 2020−300

−250

−200

−150

−100

−50

0

29oN

30oN

31oN

32oN

33oN

h’ k

h’ o

[cm

2]

(c) h’kh’

o (<0 only) vs. Southernmost Latitude of Kuroshio

Figure 8: (a) Time series of SSH anomalies averaged at southeast of Kyushu within a 4◦

width from 29◦N to 32◦N. Only negative anomalies denoting the Kuroshio trigger meandersare shown. (b) Time series of SSH anomalies averaged in the 4◦ band east of the RyukyuIslands from 24◦N to 29◦N. (c) Product of the SSH anomalies shown in (a) and (b) with a lagof 4 months. Only negative anomalies denoting trigger meanders followed by positive SSHanomalies east the Ryukyu Islands are shown. Black line denotes Kuroshio’s southernmostlatitude south of Japan.

41D

ep

th [

db

]

(a) T(y,z): KS1702

28°N 30°N 32°N 34°N

1400

1200

1000

800

600

400

200

0(b) S(y,z)

28°N 30°N 32°N 34°N

Temperature [oC]

2 4 6 8 10 12 14 16 18 20

De

pth

[d

b]

(d) T(y,z): RF1802

28°N 30°N 32°N 34°N

1400

1200

1000

800

600

400

200

0

Salinity [PSU]33.5 34 34.5 35

(e) S(y,z)

28°N 30°N 32°N 34°N

(c) Ug(y,z)

28°N 30°N 32°N 34°N

−100 −50 0 50 100

(f) Ug(y,z)

Ug [cm/s]

28°N 30°N 32°N 34°N

Figure 9: Cross section profiles of (a) temperature, (b) salinity, and (c) geostrophic flows(referenced to 2,000 db) from the JMA cruise KS1702 of February 28–March 11, 2017. (d)–(f): Same as (a)–(c), but from the JMA cruise RF1802 of March 2–12, 2018.

42

−20 −12 −4 4 12 20

(a)

Q’ Anomaly [W/m2]

ECMWF ERA5: 09/2017−08/2019

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

−0.08 −0.048 −0.016 0.016 0.048 0.08

(b)

σBI

[1/day]

140°E 160°E 180° 160°W 140°W 120°W

−1.5 −0.9 −0.3 0.3 0.9 1.5

(d)

v’T’ [oCm/s]

140°E 160°E 180° 160°W 140°W 120°W

−1 −0.6 −0.2 0.2 0.6 1

(d)

Ekman Pumping Anomaly [10−6

m/s]

140°E 160°E 180° 160°W 140°W 120°W

Figure 10: Distributions of anomalous (a) surface turbulent heat flux, (b) maximum Eadygrowth rate at 700–850 hPa level, (c) 2∼ 8-day band-passed meridional transient eddytemperature heat flux at 850-hPa level, and (d) Ekman pumping velocity fields averagedin the recent Kuroshio large meander period of 09/2017–08/2019. With the mean SSHanomalies during this period equal to 0.05m (see Fig. 4b), the color scale for each quantitycan be converted to that shown in Fig. 5 by multiplying by 20. For example, a 10Wm−2

anomaly in (a) corresponds to a 200Wm−2m−1 regression anomaly depicted in Fig. 5a.

43

−1.5

−0.9

−0.3

0.3

0.9

1.5

SS

Ha R

egre

ssio

n [m

/m]

(a) AVISO SSHa: 01/1993−12/2019, to KEI

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

−10

−6

−2

2

6

10

Q’ R

egre

ssio

n [W

/m2/d

eg]

(b) ECMWF ERA5: 01/1979−12/2019 to KPI

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

Figure 11: (a) Regression map between the KE index and the North Pacific SSH anomalyfield in 1993–2019. The dashed box denotes the region where the SSH anomalies are usedto define the KE index shown in Fig. 4b. (b) Regression coefficient between the surfaceturbulent heat flux anomaly field and the Kuroshio path index of Fig. 12a with the formerlagging the latter by 2 months. Thick black contour denotes the time-mean wEk = 0 line.

44

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 −2°

−1°

0°

1°

2°

3°

KP

I [lat deg]

(a) Kuroshio Path Index

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020−3

−1

1

3

PD

O [

oC

]

(b) PDO

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020−10

−6

−2

2 6

10

PM

M [

oC

]

(c) PMM

Figure 12: (a) Time series of Kuroshio path anomalies measured from thesouthern coast of Japan. (b) Time series of the PDO index provided byhttp://research.jisao.washington.edu/data sets/pdo. (c) Time series of the PMM indexprovided by http://www.aos.wisc.edu/∼dvimont/MModes/Data.html. In all figures, thickblack lines denote the time series after a low-pass filter that has a half-power at 6 months.

45

−1

−0.5

0

0.5

1

Wek R

egre

ssio

n [10

−7m

/s/o

C]

(a) Regressed to PMM: 01/1979−07/2017

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

−2

−1

0

1

2

Wek R

egre

ssio

n [10

−7m

/s/o

C]

(b) Regressed to PDO: 01/1979−07/2017

140°E 160°E 180° 160°W 140°W 120°W 20°N

30°N

40°N

50°N

60°N

Figure 13: Distributions of the wEk anomalies regressed to (a) the PMM index of Fig. 12cand (b) the PDO index of Fig. 12b for the period of 01/1979–07/2017. Thick black contourin both figures denotes the time-mean wEk = 0 line.